This report describes a series of six experiments designed to evaluate the performance of programmers using different forms of software documentation. The results from the first five experiments suggested that for sequential programs, succinct symbology leads to better performance than English prose, and a branching spatial arrangement provides the clearest display of control flow. The final experiment suggested that the most appropriate type of documentation for concurrent programs is different than the most appropriate type of documentation for strictly sequential programs. For simple programs and simple modifications, it was not crucial whether interprocess communications or control-flow information was highlighted in the documentation format. For more complex problems, it appeared that control-flow information was not necessary, and, in fact, may have interfered with performance. This report also describes the contents of the book: Human Factors in Computer Systems. This book is being published by Ablex Publishers and it contains several papers selected from the conference held in Gaithersburg, Maryland in March 1982. (Author)

plus-circle Add Review

comment

Reviews